1. Global sea-surface waves
Dr Lucy Bricheno, Marine Systems Modelling (Dept. Science and Technology), National Oceanography Centre
This animation shows how wind driven ocean waves grow and move over the sea surface. The 'hot' colours red and orange show high waves, and areas with 'cold' colours (white and blue) are calm. Waves grow where the winds are strongest, generating long 'swell' waves which move across the ocean until reaching land. Large, periodic storms are seen to grow and dissipate in the North Pacific, North Atlantic, and Southern Ocean. The strongest storms and highest waves are seen in the winter.
At the poles, sea-ice grows in winter, this stops waves growing, as the wind cannot pass energy to the surface water, so waves are suppressed. In the Southern Ocean, the waters circulating around Antarctica, there is a stretch of water that is totally uninterrupted by land. Here, waves can grow very large without losing energy, so some of the largest waves in the world are observed.
2. Modelling hydrogen deposition: assessing climate impacts of interactive methane and hydrogen
Dr Megan Brown, Yufus Hamied Department of Chemistry, University of Cambridge
Atmospheric hydrogen indirectly contributes to climate change by extending the lifetime of methane and increases the production of other greenhouse gases. Hydrogen lifetime in the atmosphere is poorly constrained; two main ways it is destroyed is by chemical reactions (30%) and soil uptake from microbes (70%). The soil uptake has a large uncertainty (+/-60%), and is important for determining the proportion of hydrogen emissions which cause chemical feedbacks impacting methane. We implemented a two-layer soil deposition scheme into a global climate model (Unified Model) to capture the seasonal and global distribution of atmospheric hydrogen. This figure shows the global hydrogen uptake from the final year of a 20-year simulation which has been run on ARCHER2, with an underlayer of topography. We've combined this soil uptake scheme with interactive methane to further understand the impact of hydrogen on methane and its potential greenhouse gas effects.
3. Dramatic ionization from the nitrogen dimer during ultrafast laser irradiation
Mr Dale Hughes, Centre for Light-Matter Interactions, School of Maths and Physics, Queen's University Belfast
Using the ARCHER2 compute capabilities, we have carried out precision calculations to ascertain the behavior of a Nitrogen molecule subject to intense laser fields, on a timescale on the order of femtoseconds. In these calculations, we directly calculate the motion of the electrons in the molecule. The resulting density of charge for each electron in the system can then be visualized using modern rendering techniques, originally developed for the film industry by DreamWorks, and also rendered using ARCHER2 compute. Here, charge associated with two of the electrons are rendered in red and blue. Such pure and intense irradiation, over such short timescales, is not seen in nature, so the visualisations are a testament to humanity's recent progress on both laser and computer technologies, and our deepening understanding of quantum processes.
4. Pulsed-power driven magnetic reconnection
Dr Nikita Chaturvedi, Department of Physics, Imperial College London
Magnetic reconnection occurs when anti-parallel magnetic field lines undergo a rapid change in topology. This is thought to occur in the collisional, radiative regime in black hole accretion disks. This simulation is of 'exploding wire arrays' driven by Z, the world's largest pulsed-power facility, used to recreate astrophysical conditions in the laboratory. An immense, fast rising current (26 million amperes in 300 nanoseconds) is passed through a cage of wires, which rapidly heat and form a surface plasma that is accelerated towards plasma from an adjacent array. The ablation streams carry azimuthal magnetic fields, enabling reconnection to occur in the interaction region. The central slice through the simulation shows plasmoids (flux-ropes) forming in the reconnection region that go MHD unstable and are advected with the outflows. This simulation was performed using the CHIMERA radiation-magnetohydrodynamics code with 4 billion computational cells across 16,000 cores on ARCHER2.
5. Icequakes: vibrating supercooled water in the nanopore
Pengxu Chen, School of Engineering, The University of Edinburgh
Ice nucleation within supercooled liquid is the process where the first few ice crystals, comprising a few molecules, begin to form. The accompanying image vividly illustrates ice (represented by white molecules) nucleating from supercooled water (depicted by translucent blue molecules) confined within a nanopore with a vibrating wall, a system of interest to confined nanopores. My results reveal that negative pressure, generated as the vibrating surface stretches the supercooled water, serves as a catalyst for ice nucleation by encouraging the formation of larger ice-like clusters. Once sufficient large clusters form and accumulate, the critical threshold is reached, and an ice front forms that rapidly envelopes the entire domain. This work was performed using molecular dynamics simulations in LAMMPS on ARCHER2.
6. Icequakes: vibrating supercooled water in the nanopore
Pengxu Chen, School of Engineering, The University of Edinburgh
Ice nucleation within supercooled liquid is the process where the first few ice crystals, comprising a few molecules, begin to form. The accompanying vidoe vividly illustrates ice (represented by white molecules) nucleating from supercooled water (depicted by translucent blue molecules) confined within a nanopore with a vibrating wall, a system of interest to confined nanopores. My results reveal that negative pressure, generated as the vibrating surface stretches the supercooled water, serves as a catalyst for ice nucleation by encouraging the formation of larger ice-like clusters. Once sufficient large clusters form and accumulate, the critical threshold is reached, and an ice front forms that rapidly envelopes the entire domain. This work was performed using molecular dynamics simulations in LAMMPS on ARCHER2.
7. Separated vortex ring above a three-dimensional porous disc
* * * Winning Early Career entry * * *
Dr Chandan Bose, Aerospace Engineering, School of Metallurgy and Materials, University of Birmingham
The primary motivation of studying the wake of a permeable disc stems from the idea of uncovering the role of porosity and permeability on the wind dispersal of biological seeds, such as dandelions. A stable separated vortex ring (SVR) forms in the wake of dandelion seeds, promoted by a filamentous pappus at the head of the seed, having porosity > 0.9. An SVR with a similar topology is also observed behind permeable disks with uniform permeability and porosity. Its existence is, however, limited within a specific range of the Reynolds number, and the Darcy number. The SVR is a stable toroidal vortex, whose closed streamlines are separated from the solid body. This flow feature is correlated with the high drag per projected area of the dandelion, that is an order of magnitude higher than that of an impervious disk with an equal area.
08. Leading-edge vortex instability over a plunging swept wing
Alexander Cavanagh, Autonomous Systems and Connectivity, University of Glasgow
The image shows the flow field over a plunging swept wing and the vortices that are formed due to the motion. The colouring is by normalised velocity magnitude. Instabilities are present where the leading-edge vortex is attached to the wing surface, which cause the characterstic shape of the vortex. Other features include the tip vortex and the trailing-edge vortex.
This research focusses on investigating the unstable nature of these vortices over swept wing geometries and how this affects lift produced by small fliers, known as micro-air vehicles. To provide context, the wing shown has a physical size of 0.24m wing span and 0.08m chord length.
09. Instantaneous spatial structure of elastic turbulence at vanishing Reynolds number and large Weissenberg number
Martin Lellep, School of Physics and Astronomy, The University of Edinburgh
Viscoelastic fluids like dilute polymer solutions belong to the unexplored class of non-Newtonian fluids & exhibit chaotic motion at vanishing Reynolds numbers (often called elastic turbulence), which is in stark contrast to their Newtonian counterpart. We have created the first-ever 3D simulations of viscoelastic fluids using up to 32,000 CPU cores on ARCHER2. Those simulations provide an additional pillar next to experiment and theory to elucidate the exiting but unexplored unsteady state dynamics of viscoelastic fluids. For example, the simulations have provided evidence for chaotic motion of viscoelastic fluids being organised by exact coherent states, a theoretical concept that helped to understand the Newtonian counterpart. The figure presents simulation data by showing instantaneous vertical slices of the normalised polymer elongation at several positions in the 3D volume.
10. Characterisation of the electron distribution of 2D materials and their adsorbates
Dr Zhongze Bai, Mechanical Engineering, University College London
Electron Localization Function (ELF) and Charge Density Difference (CDD) are important tools in quantum chemistry to analyse electron distribution, bonding properties and intermolecular interactions, which aid in studying chemical reactions, material properties, and electronic stability. In our work, we carried out electronic analysis for CuNC-4-pyridine (a Cu-base single-atom catalyst) with and without COOH adsorption (one of the key intermediates during CO2 electroreduction). In detail, ELF and CDD analysis present the bonding nature of CuNC-4-pyridine. And CDD of adsorbed COOH on CuNC-4-pyridine is used to reveal the CO2 reduction characteristics on CuNC-4-pyridine at the electronic level. Such information is helpful for understanding electronic structures, bonding and reactivity of CuNC-4-pyridine, providing valuable insights for catalyst design, reaction mechanisms, determination of the stability of catalysts and selection of the adsorption sites for intermediates. The quantum chemical calculations were performed on ARCHER2.
11. Effects of "wake steering" on wind turbine flows. The upwind turbine on the right is rotated to steer its wake away from the downstream turbine increasing the overall efficiency.
* * * Winning Video * * *
Dr Sébastien Lemaire, EPCC, University of Edinburgh
The video shows two computational fluid dynamics simulations of wind turbine flows ran using XCompact3D. On the left of the video, two turbines are placed are directly in line, while the second simulation, on the right, shows the angle of the front turbine "steered" by 24 degrees. The aim of "wake steering" is to redirect the wake from the turbine at the front and thus its impact on the turbine that sits downstream to increase the overall power output, in this case by 12%.
The video was generated using Blender in a photorealistic way to appeal to a broader audience. The normally invisible wakes are represented using Q-criterion, a method that highlights vortices from flows. It highlights the scale of off-shore wind turbines and their wakes, the slow evolution of the produced flow and the impact of "wake steering".
The simulations were performed by Andrew Mole (https://arxiv.org/abs/2407.20832).
12. Molecular dynamics computation of Glycerol monooleate (GMO) molecules gathering around water droplets to form reverse-micelles helping lubrication efficiency
Dr Sébastien Lemaire, EPCC, University of Edinburgh
Lubrication is a complex but important problem; its study can help reduce wear between solid surfaces. Glycerol monooleate (GMO) is known to have impressive lubricating properties. The addition of a small amount of GMO to a lubricant can drastically lower its friction. The GMO molecules gather around small droplets of water forming "reverse-micelles". Under shear, these reverse-micelles get caught on the roughest areas of the solid surface and reduce the resultant friction. The image, from a LAMMPS computation, shows the water molecules in blue, GMO molecules as their atomistic components (black, red and purple) while the solvent is hidden for clarity. The visualisation was done using Blender, allowing for fine control over camera placement and lighting, which helps highlight important features of the computation while producing a visually pleasing image suitable for non-scientific audiences.
The dataset and domain knowledge were provided by Rui Apóstolo (10.1039/d3nr05080g).
13. 3D streamlines around porous disks at varying incidence angles
Doudou Huang, Institute for Energy Systems, School of Engineering, The University of Edinburgh
We numerically investigate the flow around three-dimensional (3D) porous disks at various incidence angles. Computational Fluid Dynamics (CFD) simulations are conducted on ARCHER2 (project: e812) to study how inclination affects the aerodynamics of porous disks. The image presents 3D streamlines of the flow around high-permeability porous disks at different incidence angles, coloured by streamwise velocity.
14. Large-eddy simulation of a diurnal cycle of the urban boundary layer over London
Sam Owens, Department of Civil and Environmental Engineering, Imperial College London
Earth's atmosphere goes through daily cycles due to the surface being heated by the sun during the day and releasing stored energy at night. This video shows a 3-hour period in the early afternoon, taken from a 24-hour simulation of the atmosphere up to 1 km above central London. The tall building visible is the BT tower, giving a sense of scale to the air flow, which is modelled down to a resolution of 2 metres. The visualisation shows how air temperature varies, adjusted for the fact that pressure decreases with height. The behaviour is dynamic and complex; as the surface heats up, warm air rises, creating a turbulent lower layer that grows and mixes with the calmer air above. High-resolution simulations like this are essential for studying how the built environment interacts with the local climate, helping to make cities more resilient, sustainable and adaptive to future environmental challenges.
15. Exploring an exploding nanodroplet
Dr Ruitian He, Department of mechanical engineering, University College London
Droplet explosion, which fragments the droplet into several secondary droplets, is generally considered to improve the homogeneity of atom distribution and benefit the atomization process. However, the underlying physics within the droplet, represented in grey in this image, remains unclear. Through reactive molecular dynamics simulations in LAMMPS on ARCHER2, we uncovered the explosion mechanism of a nanometer-sized metal-containing droplet at the atomistic scale. This image illustrates the reaction pathways of molecular cluster, along with cluster analysis and energy profiles of the nanodroplet. This image was generated using Ovitio post-processing software.
16. A 'shaking' droplet: mapping lithium atom trajectory
Dr Ruitian He, Department of mechanical engineering, University College London
This image depicts a slice of a 'shaking' droplet, showing large fluctuations due to atom motion and gas release during the droplet explosion. We conducted reactive molecular dynamics simulations using LAMMPS to visualize the atom dynamics within the droplet on ARCHER2. The streamline in this image represents the trajectory of lithium atoms over 1 femtosecond, demonstrating rigorous motion of lithium atoms during the explosion. This image was generated using Ovitio post-processing software.
17. The shape of water
Eric Breard, School of Geoscience, University of Edinburgh
The still image from the video shows "tsunami" waves propagating in a shallow water tank, 1.8 m in length, as a result of a gas-particle collapse. This research is part of "Boiling Point," a collaboration between the Geophysical Flow Lab at the University of Edinburgh and Olivier Desjardins at Cornell University. The aim of this research is to simulate three-phase flows to understand how volcanic flows can generate multi-hazards, including tsunami generation caused by landslides and pyroclastic flows entering water bodies. The simulation uses Desjardins' multiphase flow solver, NGA2, a high-fidelity Euler-Lagrange solver where fluid-particle interactions are two-way coupled (solving Navier-Stokes equations), and particle behaviour is resolved using the discrete-element method (DEM). In this study, our first goal is to validate our simulations by reproducing laboratory experiments and understanding the "Shape of Water" waves before progressing to more complex scenarios involving heat and phase change.
18. Turbulent times
* * * Winning Image and overall competition winning entry * * *
Eric Breard, School of Geoscience, University of Edinburgh
The image shows a reproduction of the June 3rd, 2018 eruption at Fuego volcano, Guatemala. The volcanic plume reached around 15 km high, depositing ash to the north. The large-eddy simulation, which included the wind field, used the MFIX-classic Eulerian (solid-gas) solver and ran on ARCHER2's 1,200 CPU cores for 40 hours (simulating 2,000 seconds of real time). A North-South (XY plane) slice shows the log10 of particle concentration, highlighting entrainment's effect on plume dilution and buoyancy. Sediment waves produced ash fall on the nearby Acatenango volcano. The topography was modeled using a Lidar digital elevation model, resampled to a 20-meter resolution. This work is a collaboration between the Geophysical Flow Lab at the University of Edinburgh, Dr Jordan Musser (NETL, US DOE), and Pr Joe Dufek (University of Oregon), aiming to reproduce complex volcanic plumes under real atmospheric conditions, including 3D wind fields.
19. Growth of hexagonal boron nitride
Anthony Payne, School of Chemistry and Chemical Engineering, University of Surrey
Hexagonal boron nitride (hBN) can be synthesised via chemical vapour deposition (CVD) of borazine (B3N3H6) on a ruthenium surface. The growth process, intermediates and resulting structures depend highly on temperature and precursor exposure. We developed a first-principles microkinetic model that accurately predicts the temperature dependence of hBN growth from initial adsorption to nanoporous intermediates and final overlayer. Adsorption at low temperatures (green) is followed by sequential partial dehydrogenation (purple/blue) and diffusion with increasing temperature. Sufficient exposure leads to dimerisation and then polymerisation into nanoporous intermediates (red), while excess exposure results in disordered phases (grey). Complete dehydrogenation at high temperatures leads to hBN formation (yellow). The model aligns with experimental data, providing insights into atomic structure and synthesis. The animation was created using structures calculated from Density Functional Theory via CASTEP on ARCHER2 (excluding the inferred disordered phase) before combination and animation in Blender to highlight each phase.
20. Simulation of a heartbeat in the cardiovascular system through multi-component coupling
Sharp Lo, Centre for Computational Science, University College London
This video showcases a successful simulation of a heartbeat and blood flow in the vasculature, achieved by coupling the 3D electromechanical heart model in Alya with the 3D vascular flow model in HemeLB. These models are developed by separate research groups (BSC and UCL), focus on different dynamical scales (macroscopic and mesoscopic), and employ distinct discretisation methods (finite element and lattice Boltzmann). The simulation was executed on ARCHER2 using 4096 cores across 32 nodes. This successful collaboration highlights the potential of multi-component coupling and represents a significant milestone towards creating a virtual human.
21. Active turbulence in microswimmer suspension.
Dr Giuseppe Negro, ICMCS, School of Physics and Astronomy, University of Edinburgh
The colour represents the largest eigenvalue of the Q-tensor, that gives the degree of local microorganism alignment. Resolution: 512 Fourier modes in each spatial direction comprising in total 1.7*E+9 degrees of freedom.
The coordinated behaviour of living organisms, like the collective motion of bird flocks, has long fascinated scientists. These behaviours spurred the development of the new "active matter" field in physics. Active matter systems, like bacterial suspensions, use internal energy to move, producing chaotic patterns like jets and vortices at high densities. This phenomenon, called “active turbulence”, differs from traditional turbulence in Newtonian fluids and remains poorly understood.
Our research aims to quantify active turbulence by treating it as a non-equilibrium phase transition. We developed a 3D model with 13 spatial fields and up to 10 billion degrees of freedom, whose dynamics are numerically solved using Dedalus, a Python-based framework for pseudo-spectral methods. The Figure shows a 3D configuration, where the colour corresponds to the Q-tensor’s largest eigenvalue, representing microorganism local alignment. Using 512 Fourier modes for each dimension, this simulation ran on ARCHER2 for 48 hours on 128 nodes.
22. Active turbulence in 2D microswimmer suspension.
Dr Giuseppe Negro, ICMCS, School of Physics and Astronomy, University of Edinburgh
The colour represents the largest eigenvalue of the Q-tensor, that gives the degree of local microorganism alignment.
The coordinated behaviour of living organisms, like the collective motion of bird flocks, has long fascinated scientists. These behaviours spurred the development of the new "active matter" field in physics. Active matter systems, like bacterial suspensions, use internal energy to move, producing chaotic patterns like jets and vortices at high densities. This phenomenon, called “active turbulence”, differs from traditional turbulence in Newtonian fluids and remains poorly understood.
Our research aims to quantify active turbulence by treating it as a non-equilibrium phase transition. We developed a 2D model, whose dynamics is numerically solved using Dedalus, a Python-based framework for pseudo-spectral methods. The movie shows the time evolution of the system in the turbulent regime, where the colour corresponds to the Q-tensor’s largest eigenvalue, representing microorganism local alignment. Using 3000 Fourier modes for each dimension, this simulation ran on ARCHER2 for 48 hours on 50 nodes.